Mining and corruption

Abstract

This study examines the links between mineral dependency and corruption. Specifically, it develops a cross-section econometric model that estimates the effects of per capita income, fuel exports, non-fuel exports, and average per unit value of mineral exports on corruption.

The results indicate that fuel and non-fuel mineral exports affect corruption differently. Corruption increases with fuel exports in an unambiguous manner. Non-fuel mineral exports tend to increase corruption only in poor countries, particularly affecting those exporting high value mineral commodities such as diamonds and gold. In richer countries, non-fuel mineral exports actually reduce corruption. In addition, while economic development reflected by rising per capita income ultimately reduces corruption, the evidence indicates that at early stages of development, advances in per capita income are associated with greater corruption. These findings suggest that the links between corruption on the one hand and mineral dependency and per capita income on the other are far more complex than widely recognized.

Introduction

One might expect mineral wealth, like other forms of wealth, to contribute positively to economic development and growth. Yet, over the past two decades a number of scholars have found that countries heavily dependent on mining and mineral exports tend to grow relatively slow. This literature has led some to conclude that mineral wealth, far from being a benefit, is actually a curse for many developing countries. Among the many possible explanations for this troubling and counter-intuitive finding is the possibility that mineral wealth fosters corruption.¹

Corruption is often defined as the use of public goods for private benefit. A good illustration is the government official who accepts bribes to grant licenses and permissions, or to inhibit the entry of new competitors. But corruption can also exist in the private sector, for example, when an employee of a private firm uses the firm's resources for his or her own benefit.

While corruption may occasionally have positive effects, such as circumventing bureaucratic constraints, normally it retards economic growth, as Bardhan (1997), Poirson (1998), and others have shown. As a result, international organizations have tried to help poor countries struggling with corruption to understand its causes (Werlin, 2005).

While a considerable amount of literature does exist on corruption,² most of the available studies are either theoretical in nature or case studies. Only over the past decade or so have data on corruption—or to be precise, perceptions of corruption—become available for cross-country empirical studies.³ The latter assess the impact on corruption of a wide range of economic and cultural variables in a variety of different models. While their conclusions vary, a persistent finding is that per capita income by itself can account for a large percentage of the differences in corruption among countries. This suggests that per capita income is a good proxy for income inequality, democracy, trade openness, education, culture, and the host of other variables assumed to affect corruption (Paldam, 2002).

To our knowledge, only four of the available empirical studies consider the impact of natural resource dependency on corruption:

• Leite and Weidmann (1999): This work identifies four categories of natural resources—fuels, non-fuel minerals, food, and agricultural raw materials—and finds that the fuels and non-fuel minerals promote corruption, while food and agricultural raw materials do not.⁴ It concludes that the effects of fuel and non-fuel minerals on corruption are the same, since both sectors, unlike food and agricultural raw materials, have high capital–labor ratios. It also considers the effects of corruption and natural resource dependency on economic growth, and finds that both hinder growth, with natural resource dependency doing so even after controlling for corruption.

• Petrovsky (2004): This MS thesis estimates corruption among countries as a function of a number of variables, including democracy, income inequality, per capita GDP, schooling, and mineral resource dependency. The latter is assessed in separate equations by the ratio of fuel exports to GDP, the ratio of metal and ore exports to GDP, and the ratio of total mineral exports to GDP. Petrovsky includes interactive terms between mineral resource dependency and the other explanatory variables to take account of the possibility that the effect on corruption of mineral resource dependence may vary depending on the level of education, the strength of democratic institutions, the per capita GDP, and other explanatory variables. Although his results suffer from multicollinearity, he does find that dependency on metals and ores (though unlike Leite and Weidmann not fuels or total minerals) significantly increases corruption, but interestingly only in countries with a narrow tax base and low levels of education.

• Douoguih (2005): This cross-section study of more than 150 countries specifies a model that makes corruption a function of per capita GDP, the ratio of fuel exports to GDP, the ratio of non-fuel exports other than diamonds to GDP, and the ratio of diamond exports to GDP. The study finds that a high ratio of fuel exports to GDP (or of fuel exports to total exports) does significantly increase corruption, a finding consistent with Leite and Weidmann but not Petrovsky. It also finds that a high ratio of non-fuel exports other than diamonds to GDP (or of non-fuel exports other than diamonds to total exports) tends to reduce corruption. Finally, it finds that a high ratio of diamond exports to GDP also reduces corruption, but this effect is not statistically significant. Since the analysis does not take into account possible interactive effects among the explanatory variables, we do not know whether the effects on corruption of the fuel, non-fuel mineral, and diamond variables vary with the level of per capita income.

• Dietz et al. (2007): This panel data study uses a dynamic Arellano–Bond model to explain lack of corruption as a function of various variables (gross national income per capita, GDP growth, age distribution of the population, urbanization and institutional quality) and resources dependency (measured by the ratio of resource exports to total exports). Resource exports include fuels, non-fuel mineral commodities, and agricultural raw materials, though no attempt is made to identify the separate influence of these three groups of resources. The study finds a positive and significant relationship between lack of corruption and resource exports and concludes that it is ultimately policy failure, not the exploitation of natural resources, that fosters corruption.

Douoguih, by assessing separately the effects of diamonds and other non-fuel minerals, raises the possibility that within the non-fuel minerals group differences may exist between the impact on corruption of high value commodities, such as gold, platinum, and diamonds, and the impact of lower value commodities, such as iron ore, bauxite, and copper. While the production and export of non-fuel minerals are often associated with corruption, almost always the examples cited are high value commodities—whose exploitation often generates sizable economic rents—and the countries that produce them, such as Angola, Sierra Leone, and the Democratic Republic of the Congo (Good, 1994; Sherman, 2000; Diamond Industry Annual Review, 2004; Gberie, 2002). Indeed, concern over the use of revenues from diamond mining has led to the creation of a certification system, called the Kimberly Process, to ensure diamond revenues are not used to support conflict or to pay for weapons, particularly in Africa.

There are also good conceptual reasons to believe that the high value commodities among the non-fuel minerals may promote corruption more than those of lower value. Gold, platinum, and diamonds are easier to hide and to smuggle. The rents their production creates are more concentrated, making it easier for rent-diverting activities to succeed, including those involving corruption.

This study, which is based on Petermann (2005), proposes to assess empirically the effects on corruption of fuel and non-fuel mineral exports, including the possibility that these effects differ for high value and low value products. To do so, it develops a cross-section econometric model that relates corruption with various explanatory variables, including fuel and non-fuel mineral exports. Like Petrovsky, it will also take account of possible interactive effects between the explanatory variables.

The presentation is organized as follows. The section “Model” describes the model. The sections “Estimation and results” and “Per capita income thresholds for PCYi, OREi, and APi” examine the results and discuss their implications. The section “Conclusions” summarizes the conclusions while highlighting several caveats.